用量子化学参数预测烯烃聚合物介电常数  

Prediction of Dielectric Constants for Polymer by Quantum Chemical Descriptor

作　　者：禹新良[1] 张仕华[2] 王学业[3] 

机构地区：[1]湖南工程学院化学化工学院,湘潭411104 [2]湖南工程学院网络信息中心,湘潭411104 [3]湘潭大学化学学院,湘潭411105

出　　处：《湖南工程学院学报（自然科学版）》2008年第4期65-68,共4页Journal of Hunan Institute of Engineering(Natural Science Edition)

基　　金：湖南省教育厅科研资助项目(07C205);湖南工程学院科研资助项目(0761)

摘　　要：在B3LYP/6-31G(d)水平上对烯烃聚合物单体进行密度泛函理论计算,得到4个量子化学参数,即分子偶极矩(μ),分子中最低未占分子轨道能级(ELOMO),熵(S)及原子最负电荷q-,并用来预测烯烃聚合物介电常数(ε).误差反向传播的人工神经网络方法(BPANN)用来拟合介电常数与这4个参数可能存在的非线性关系.根据调整参数得到最佳网络模型结构为[4-1-1],模拟值与实验值非常接近,训练集与测试集的均方根(rms)误差分别为0.430和0.321.Density functional theory （DFT） calculations are carried out in the prediction of dielectric constants （ε） of vinyl polymers at the B3LYP/6-31G（d） level. 4 quantum chemical descriptors calculated from monomers of polymers, the dipole moment μ the energy of the lowest unoccupied molecular orbital ELUMO, the entropy S, and the most negative net charge of atom q- are used to build a general quantitative structure-property relationship （QSPR） model for ε. The typical back-propagation （BP） neural network is employed for fitting the possible non-linear relationship existing between the four descriptors and ε. The optimal condition of the neural network is obtained by adjusting various parameter. Simulated with the final optimum BP neural network^[4-1-1], the results show that the fitted e values are in agreement the experimental data, the root mean square error （rms） being 0. 430 for the training set and 0. 321 for the test set.

关 键 词：密度泛涵理论 介电常数 聚合物 结构-性能定量关系 量子化学参数 

分 类 号：O6-04[理学—化学] O631